Signed-off-by: Vassil Vassilev <v.g.vassilev@gmail.com>

This is preferred way, while correctly handles object ownership

Fixing:  warning: format ‘%lu’ expects argument of type ‘long unsigned int’, but argument 2 has type ‘std::size_t {aka unsigned int}’ [-Wformat=]

One requires appropriate search method to found primitive and pad to
which it belong

When JS-side renders such objects, NDC position is calculated.
Now this position applied back to the server to preserve it.

Object will be referenced from client and updated correspondently

Used for different settings provided by JSROOT, try to avoid full
objects streaming like TFrame or TPaveText 

Also for web-based canvas we need such objects to preserve different
attributes, which can be changed by client 

will fail in exp PyROOT

Due to a missing converter in the old cppyy, one cannot call a function
that takes a `const char*` using a nullptr.
To make RooDataSet::sumEntries callable from Python, it now also takes
empty strings.

See also:
https://root-forum.cern.ch/t/sumentries-cutrange-in-pyroot/33499

Instead, test for libC++ and language version.
This fixes MacOS builds with >=C++14 and tests enabled.

The RooFit release notes were slightly outdated after the deprecation
macro R__SUGGEST_ALTERNATIVE had been merged.

This prevents the vector to be reallocated if the size exceeds the capacity of
the vector causing invalid pointer accesses.

Fixes ROOT-7749.

When a RooDataSet is constructed without a weight variable, weights are ignored when events
are added to the dataset. Now, there will be a clear warning notifying the user that an event
weight is being ignored.
The documentation has also been updated.

See also:
https://root-forum.cern.ch/t/fit-to-a-weighted-unbinned-data-set/33495

* Fix doxygen generation with Python 3

* Adjust tutorials for doxygen documentation generation.
When run inside the makeimage.py wrapper during doxygen generation the
variable scope is different than when the tutorials are run stand-alone.
The change in tutorials/math/tStudent.py is Python 3 specific and is due to
that list comprehensions in Python 3 are run in an implicit function scope.

The new release includes some improvements in both Forward and
Reverse mode:
* Extend the way to specify a dependent variables. Consider function,
  `double f(double x, double y, double z) {...}`, `clad::differentiate(f, "z")`
  is equivalent to `clad::differentiate(f, 2)`. `clad::gradient(f, "x, y")`
  differentiates with respect to `x` and `y` but not `z`. The gradient results
  are stored in a `_result` parameter in the same order as `x` and `y` were
  specified. Namely, the result of `x` is stored in `_result[0]` and the result
  of `y` in `_result[1]`. If we invert the arguments specified in the string to
  `clad::gradient(f, "y, x")` the results will be stored inversely.
* Enable recursive differentiation.
* Support single- and multi-dimensional arrays -- works for arrays with constant
  size like `double A[] = {1, 2, 3};`, `double A[3];` or `double A[1][2][3][4];`

See more at: https://github.com/vgvassilev/clad/blob/v0.5/docs/ReleaseNotes.md

- when ROOT::IsEnableImplicitMT() TMVA will run by default in MT
- when TMVA::gConfig().EnableMT(0) or TMVA::gConfig().EnableMT(nthreads > 1)  TMVA will run in MT independently of ROOT::IsEnabledImplicitMT()
- when TMVA::gConfig().EnableMT(1) or TMVA::gConfig().DisableMT()  TMVA will run sequentially and its thread pool will be released 

since it is a special class: we are not interested in the actual
layout, it is opaque for the ROOT IO being it part of the STL.

webui should always be build with root7, if other is not specified

Show correspondent status message

It should be possible build root7, but not build webui